Hepatocyte Growth Factor-regulated Tyrosine Kinase Substrate (HRS) Mediates Post-endocytic Trafficking of Protease-activated Receptor 2 and Calcitonin Receptor-like Receptor*

The E3 ligase c-Cbl ubiquitinates protease-activated receptor 2 (PAR2), which is required for post-endocytic sorting of PAR2 to lysosomes, where degradation arrests signaling. The mechanisms of post-endocytic sorting of ubiquitinated receptors are incompletely understood. Here, we investigated the role of hepatocyte growth factor-regulated tyrosine kinase substrate (HRS), in post-endocytic sorting and signaling of PAR2. In HEK-PAR2 cells, PAR2 activating peptide (PAR2-AP) induced PAR2 trafficking from the cell surface to early endosomes containing endogenous HRS, and then to lysosomes. HRS overexpression or knockdown with small interfering RNA caused formation of enlarged HRS-positive endosomes, where activated PAR2 and c-Cbl accumulated, and PAR2 failed to traffic to lysosomes. Overexpression of HRS prevented PAR2-AP-induced degradation of PAR2, as determined by Western blotting. Overexpression of HRS mutant lacking an ubiquitin-binding motif similarly caused retention of PAR2 in enlarged endosomes. Moreover, HRS overexpression or knockdown caused retention of ubiquitin-resistant PAR2Δ14K/R in enlarged HRS-containing endosomes, preventing recycling and resensitization of PAR2Δ14K/R. HRS overexpression or knockdown similarly prevented lysosomal trafficking and recycling of calcitonin receptor-like receptor, a non-ubiquitinated receptor that traffics to lysosomes after sustained activation and recycles after transient activation. Thus, HRS plays a critically important role in the post-endocytic sorting of single receptors, PAR2 and CLR, to both degradative and recycling pathways. This sorting role for HRS is independent of its ubiquitin-interacting motif, and it can regulate trafficking of both ubiquitinated and non-ubiquitinated PAR2 and non-ubiquitinated CLR. The ultimate sorting decision to degradative or recycling pathways appears to occur downstream from HRS.

Many G protein-coupled receptors (GPCRs) 2 are rapidly endocytosed after agonist binding, but the pathway of post-endocytic trafficking depends on the receptor and the nature of the stimulus. Some GPCRs are sorted to lysosomes or proteasomes, where degradation irrevocably inactivates internalized receptors and prevents uncontrolled signaling during chronic stimulation (1)(2)(3)(4). Other GPCRs recycle to the plasma membrane, which mediates resensitization of signal transduction (5)(6)(7)(8). Our understanding of the mechanisms underlying the critical sorting of GPCRs to these divergent pathways, degradative or recycling, is incomplete and controversial. Ubiquitination of certain receptors, exemplified by protease-activated receptor 2 (PAR 2 ) and the chemokine (C-X-C motif) receptor 4 (CXCR4), serves as a signal for sorting receptors into a lysosomal or proteasomal degradation pathway (4, 9 -12). However, other receptors, including the ␦-opioid receptor and calcitonin receptor-like receptor (CLR) (8,13,14), traffic to the degradative pathway by ubiquitin-independent mechanisms. It is important to elucidate the sorting machinery that diverts a receptor, in an ubiquitin-dependent or independent manner, to lysosomes or the plasma membrane, as this trafficking regulates cellular responses to agonists.
Hepatocyte growth factor-regulated tyrosine kinase substrate (HRS) is a 115-kDa endosomal protein that has attracted much interest in this context (14 -18). HRS associates with early endosomes as part of a multiprotein complex that has a key role in sorting ubiquitinated proteins to lysosomes (19 -24). It has an ubiquitin-interacting motif (UIM) and can bind directly to ubiquitinated receptors, resulting in their accumulation in internal vesicles of multivesicular bodies and their delivery to lysosomes (21,25). Non-ubiquitinated proteins are not recognized by HRS and escape sorting to the lysosomal pathway, enabling them to recycle (21,26). However, recent studies have identified additional roles of HRS in controlling trafficking on non-ubiquitinated GPCRs to degradative and recycling pathways. Thus, HRS mediates lysosomal trafficking of the non-ubiquitinated ␦-opioid receptor, and mediates recycling of the ubiquitinated ␤ 2 -adrenergic receptor and -opioid receptor (14,17). Thus, the function of HRS in post-endocytic sorting depends on the GPCR in question and its ubiquitination.
We examined the role of HRS in trafficking and signaling of PAR 2 , a GPCR that is cleaved and activated by several serine proteases that are generated during injury and inflammation (27)(28)(29). These proteases cleave PAR 2 to expose a tethered ligand domain that binds to and activates the cleaved receptor (27,28). Given the irreversible nature of proteolytic activation of PAR 2 , and its proinflammatory and nociceptive actions (29), mechanisms that arrest PAR 2 signaling are of considerable interest. Permanent signal arrest requires PAR 2 degradation in lysosomes, but the post-endocytic sorting mechanisms that target the receptor to lysosomes are incompletely understood (30,31). The E3 ubiquitin ligase c-Cbl monoubiquitinates PAR 2 at multiple sites, which is necessary for lysosomal trafficking, because a non-ubiquitinated PAR 2 mutant, PAR 2 ⌬14K/R, escapes degradation and recycles (4). Investigation of the role of HRS in post-endocytic sorting of PAR 2 to lysosomes and PAR 2 ⌬14K/R to the plasma membrane provided a unique opportunity to define the role of HRS in the post-endocytic trafficking of a single receptor with divergent fates after internalization.
For comparison, we also examined the role of HRS in the post-endocytic sorting of the receptor for calcitonin gene-related peptide (CGRP), a heterodimer of calcitonin receptor-like receptor (CLR), a GPCR, and receptor activity-modifying protein 1 (RAMP1), a protein with a single transmembrane domain. An understanding of the control of CLR and RAMP1 is important because CGRP is a potent vasodilator and mediator of neurogenic inflammation and pain transmission (32). CLR and RAMP1 recycle after transient stimulation with CGRP, but during continuous stimulation traffic to lysosomes by processes that do not require receptor ubiquitination (8). These studies enabled us to evaluate the role of HRS in the divergent trafficking of a non-ubiquitinated heterodimeric receptor to lysosomes or the plasma membrane.
cDNAs and siRNA-The pcDNA5/FRT plasmid was from Invitrogen. cDNA encoding c-Cbl with N-terminal HA11 was a gift from Dr. C. Thien (University of Western Australia) (33). Myc-tagged HRS, Myc-tagged HRS lacking the UIM (HRS⌬UIM), and Myc-tagged HRS⌬VHS (all in pcDNA3) were gifts from Dr. M. von Zastrow, Dr. J. N. Hislop, and Dr. A. C. Hanyaloglu (University of California, San Francisco, CA). cDNAs for PAR 2 , PAR 2 ⌬14K/R, CLR, and RAMP1 have been described (4,8). Knockdown of HRS using siRNA was achieved by transfection of duplex RNA oligonucleotides (Qiagen, Valencia, CA) corresponding to part of the coding region of human HRS (CGACAAGAACCCACACGTC, HRS-siRNA1), as described (14,34). To further confirm HRS knockdown specificity, a distinct siRNA directed against a second target sequence in the human HRS transcript (GCACGTCTTTCCAGAATTCAA, HRS-siRNA2) was also used (35). In control experiments, cells were transfected with nonsense duplex RNA oligonucleotide (AATTCT-CCGAACGTGTCACG).
Transfected Cells and Cell Lines-Human embryonic kidney 293 (HEK) and Henrietta Lacks (HeLa) cells were grown in Dulbecco's modified Eagle's medium containing 10% heat-inactivated fetal bovine serum in 95% air, 5% CO 2 at 37°C. The generation and maintenance of HEK-FLP cells (Invitrogen) stably expressing PAR 2 with an N-terminal FLAG epitope and a C-terminal HA11 or T7 epitope, or the ubiquitin-defective PAR 2 mutant (PAR 2 ⌬14K/R) with an N-terminal FLAG epitope and a C-terminal HA11 epitope (HEK-PAR 2 ), or CLR with an N-terminal HA11 epitope and RAMP1 with an N-terminal Myc epitope (HEK-CLR-RAMP1) have been described (4,8,36). In some experiments, HEK cells were transiently transfected using Lipofectamine TM 2000 according to the manufacturer's guidelines (Invitrogen). Cells were plated 48 h prior to the experiments. Endogenous HRS was knocked-down in HeLa cells using siRNA. Cells were plated in 6-well plates at 80% confluence. After 16 h, cells were transfected with duplex RNA strands (HRS-siRNA1, HRS-siRNA2, or control siRNA) using Lipofectamine TM 2000. After 24 h, cells were transfected again with cDNA for PAR 2 -HA11 or CLR-HA11 and RAMP1). Cells were studied 48 h after the second transfection.
Activation of PAR 2 or CLR and Drug Treatments-PAR 2 activating peptide (corresponding to the tethered ligand of rat/ mouse PAR 2 , SLIGRL-NH 2 , CPC Scientific, San Jose, CA) and rat ␣-CGRP (Bachem, Torrance, CA) were used to activate PAR 2 and CLR/RAMP1, respectively. HEK-PAR 2 or HEK-CLR-RAMP1 cells were washed three times with PBS and placed in Dulbecco's modified Eagle's medium, 0.1% bovine serum albumin. Cells were stimulated with PAR 2 -AP (100 M, various times) or CGRP (100 nM, various times). To inhibit new protein synthesis (during degradation assays and during calcium experiments), cells were incubated with cycloheximide (140 M). To interfere with Golgi vesicular transport of PAR 2 (during calcium experiments), cells were incubated with brefeldin A 1 (10 g/ml) (Sigma). Inhibitors were preincubated with cells 1 h prior to stimulation. Controls included the appropriate vehicle.
Image Analysis-Data were analyzed using Zeiss LSM 510 software. Targeting of recycling receptors to the plasma membrane was analyzed by drawing regions of interest on the outside and the inside of the plasma membrane, which allowed determination of the percentage of total cellular fluorescence at the plasma membrane, as previously described (37). Colocalization of two proteins in organelles was analyzed by drawing regions of interest around the outside of a cell in the merged image and measuring the overlap coefficient, with a coefficient of 0 indicating no colocalization and of 1 indicating complete co-localization. Ͼ20 cells were analyzed for each experiment.
Measurement of [Ca 2ϩ ] i -PAR 2 signaling was evaluated by measuring [Ca 2ϩ ] i . Cells were plated at 1.2 ϫ 10 6 cells per 35-mm dish onto coverslips coated with poly-D-lysine. Cells were washed, incubated in Hank's balanced salt solution containing Ca 2ϩ and Mg 2ϩ , 0.1% bovine serum albumin, 20 mM HEPES, pH 7.4, with 2.5 M fura-2 acetoxymethyl ester (Invitrogen) for 20 min at 37°C and washed. Fluorescence was measured at 340 and 380 nm excitation and 510 nm emission in a F-2500 spectrophotometer (Hitachi Instruments, Irvine, CA). The ratio of the fluorescence at the two excitation wavelengths, which is proportional to [Ca 2ϩ ] i , was calculated. To assess desensitization and resensitization, cells were exposed to PAR 2 -AP (100 M) or vehicle (control) for 1 h at 37°C, washed 3 times, and then challenged a second time with PAR 2 -AP (100 M) at 0 or 4 h after washing. The response to the second challenge was measured.

FIGURE 1. Localization of HRS and PAR 2 in endosomes.
A, in HEK cells, endogenous HRS colocalized with EEA1 in early endosomes (arrows). B, HEK cells were transiently transfected with Myc-HRS and were stained with Myc and HRS antibodies. Both antibodies reacted with overexpressed HRS present in enlarged vesicles (arrows) indicating specificity of the HRS antibody. Both overexpressed (Myc-HRS) and endogenous HRS colocalized with EEA1. C, HEK cells stably expressing PAR 2 were incubated with PAR 2 -AP for 0 or 30 min. Cells were fixed and proteins were localized by indirect immunofluorescence with antibodies to HA11 (PAR 2 ) and HRS. In unstimulated cells, PAR 2 was located at the plasma membrane and in a perinuclear Golgi pool, and HRS was present in intracellular vesicles (endosomes). In cells stimulated with PAR 2 -AP, PAR 2 was co-localized with endogenous HRS (arrows). Scale bars, 10 m.
Statistics-Results are expressed as mean Ϯ S.E. of n Ն 3 experiments, and were compared by Student's t test, with p Ͻ 0.05 considered to be significant. Immunofluorescence images represent n Ն 3 experiments.

PAR 2 Traffics to Early Endosomes
Containing HRS-As a first step toward examining the role of HRS in post-endocytic trafficking of PAR 2 , we localized endogenous HRS and overexpressed Myc-HRS in HEK cells by immunofluorescence, and determined whether these proteins were in endosomes. We found that in cells not overexpressing HRS, endogenous immunoreactive HRS was detected in multiple small vesicles containing EEA1, which are thus early endosomes (Fig. 1A). In cells expressing Myc-HRS, antibodies to HRS and Myc both labeled the same vesicles (Fig. 1B). As previously reported, overexpressed HRS was present in enlarged early endosomes ( Fig. 1B) (17). Thus, the HRS antibody specifically recognizes HRS, and HRS is present in early endosomes.
To assess the role of HRS in the trafficking of PAR 2 , we localized endogenous HRS in HEK-PAR 2 cells. We observed that in unstimulated cells, PAR 2 was detected at the plasma membrane and in a perinuclear location, which we have previously identified as the Golgi apparatus (30,31), whereas HRS was restricted to endosomes (21) (Fig.  1C). PAR 2 -AP (100 M, 30 min) caused internalization of PAR 2 into vesicles containing endogenous HRS (Fig. 1C). Thus, activated PAR 2 traffics to early endosomes containing HRS.
HRS Overexpression Causes Accumulation of PAR 2 and c-Cbl in Enlarged Endosomes-The overexpression of HRS disrupts function similarly to HRS knockdown, inducing formation of enlarged endosomes (14,17). Therefore, we examined the effects of overexpression of HRS on the trafficking of PAR 2 . HEK cells stably expressing FIGURE 2. Effects of HRS overexpression on trafficking of PAR 2 and c-Cbl. A, HEK-PAR 2 cells were transiently transfected with Myc-HRS and challenged with PAR 2 -AP (0, 30, or 120 min). Proteins were localized by indirect immunofluorescence using HA11 (PAR 2 ) and HRS antibodies. In unstimulated cells, PAR 2 was located at the plasma membrane (arrowheads) and in enlarged HRS-containing endosomes (arrows), which probably represents constitutively trafficked PAR 2 . In stimulated cells, PAR 2 was depleted from the plasma membrane and detected in HRS-containing endosomes (arrows). B, HEK-PAR 2 cells were transiently transfected with HA11-c-Cbl and Myc-HRS and challenged with PAR 2 -AP (0, 30, or 120 min). Proteins were localized by indirect immunofluorescence using HA11 (c-Cbl) and Myc (HRS) antibodies. In unstimulated cells, c-Cbl was cytosolic and HRS was present in intracellular vesicles (arrows). In stimulated cells, c-Cbl redistributed to enlarged HRS-containing endosomes (arrows). C, HEK-PAR 2 cells were transiently transfected with control vector (pcDNA5) or Myc-HRS, and cell-surface PAR 2 was labeled with an antibody to the extracellular epitope (Flag). Cells were challenged with PAR 2 -AP (0 or 120 min), and proteins were localized by indirect immunofluorescence using fluorescent secondary antibody to PAR 2 and HRS antibody. In unstimulated cells expressing control vector or Myc-HRS, PAR 2 was present at the plasma membrane (arrowheads). HRS overexpression caused accumulation of PAR 2 in enlarged HRScontaining endosomes (arrows). In stimulated cells, PAR 2 was depleted from the plasma membrane and detected in enlarged HRS-containing endosomes (arrows). D, HEK-PAR 2 cells were transiently transfected with control vector (pcDNA5) or Myc-HRS, and cell-surface PAR 2 labeled with an antibody to the extracellular epitope (FLAG). Cells were challenged with PAR 2 -AP (0 or 30 min), and proteins were localized by indirect immunofluorescence using fluorescent secondary antibody (PAR 2 ) and EEA1 antibody. In unstimulated cells expressing control vector, PAR 2 was present at the plasma membrane (arrowheads) and EEA1 was in intracellular vesicles (arrows). In stimulated cells, PAR 2 was detected in EEA1-containing endosomes. HRS overexpression caused accumulation of PAR 2 in enlarged EEA1-containing endosomes (arrows). E, HEK-PAR 2 cells were transiently transfected with Myc-HRS, and HRS (myc) and TfR were localized by indirect immunofluorescence. TfR, which traffics constitutively, did not accumulate in the enlarged HRS-containing endosomes (arrows). Scale bars, 10 m. PAR 2 were transiently transfected with HRS, and PAR 2 and HRS were localized by immunofluorescence. In unstimulated cells, PAR 2 was localized to the plasma membrane and also present in enlarged endosomes containing HRS ( Fig. 2A). PAR 2 -AP (30 and 120 min) caused depletion of PAR 2 from the plasma membrane and accumulation in enlarged HRS-containing endosomes (Fig. 2A).
The E3 ubiquitin ligase c-Cbl ubiquitinates PAR 2 , and PAR 2 agonists induce trafficking of c-Cbl from the cytosol to endosomes containing PAR 2 (4). Therefore, we examined the effect of HRS overexpression on trafficking of c-Cbl. In unstimulated HEK-PAR 2 cells transiently expressing c-Cbl and HRS, c-Cbl was cytosolic and HRS was present in enlarged endosomes (Fig. 2B). PAR 2 -AP caused trafficking of c-Cbl to HRS-containing endosomes (Fig. 2B). Thus, activated PAR 2 together with c-Cbl traffic to HRS-containing vesicles.
The observation that HRS overexpression causes accumulation of unstimulated PAR 2 in endosomes suggests constitutive endocytosis of this receptor. To examine this possibility in more detail, and to investigate the effect of HRS expression on trafficking of PAR 2 from the plasma membrane to endosomes, we labeled cell-surface PAR 2 using an antibody to an extracellular epitope tag (FLAG), and localized EEA1 and endogenous HRS and overexpressed HRS by immunofluorescence. In unstimulated HEK-PAR 2 cells transfected with control vector, PAR 2 was present at the plasma membrane and endogenous HRS and EEA1 were present in endosomes (Fig. 2, C and D). PAR 2 -AP (30 min) induced endocytosis of PAR 2 and its accumulation in early endosomes (Fig. 2D) as described (38). Thus, antibodytagged PAR 2 internalizes and traffics similarly to the untagged receptor. In unstimulated cells transfected with HRS, PAR 2 was present at the cell surface and also in HRS-containing vesicles (Fig. 2C), suggesting that constitutively endocytosed PAR 2 accumulates in HRScontaining vesicles (Fig. 2C). PAR 2 -AP (30 and 120 min) caused PAR 2 to traffic from the cell surface to enlarged HRS-containing, EEA1positive endosomes (Fig. 2, C and  D). Thus, overexpression of HRS does not affect the trafficking of PAR 2 from the cell surface to EEA1-containing endosomes, but causes retention of PAR 2 in these endosomes for prolonged periods. In contrast, overexpression of HRS did not affect the subcellular localization of the TfR (Fig. 2E), suggesting that the accumulation of PAR 2 in enlarged endosomes is not due to a general disruption of the endocytic pathway.
HRS Overexpression Prevents Lysosomal Targeting and Degradation of PAR 2 -Overexpression of HRS caused retention of PAR 2 in endosomes at times (120 min) when the receptor is usually in lysosomes (4), suggesting that HRS mediates lysosomal trafficking of this receptor. To examine this possibility, we studied lysosomal targeting of antibody-tagged PAR 2 in HEK- FIGURE 3. Lysosomal targeting and degradation of PAR 2 . A and B, HEK-PAR 2 cells were transiently transfected with control vector or Myc-HRS, and cell-surface PAR 2 was labeled with an antibody to the extracellular epitope (Flag). Cells were challenged with PAR 2 -AP (120 min), and proteins were localized by indirect immunofluorescence using fluorescent secondary antibody (PAR 2 ) and EEA1 and LAMP1 antibodies. In cells expressing control vector, PAR 2 did not colocalize with EEA1 but did colocalize with LAMP1. In cells overexpressing HRS, PAR 2 colocalized with EEA1 in enlarged endosomes, but did not colocalize with LAMP1. Scale bars, 10 m. C, quantification of the effect of HRS overexpression on PAR 2 trafficking to lysosomes. The overlap coefficient between PAR 2 and EEA1 or PAR 2 and LAMP1 was determined (0, no overlap; 1, complete overlap) in HEK-PAR 2 cells transiently transfected with control vector or Myc-HRS. n, 20 to 28 cells each. *, p Ͻ 0.05. D, HEK-PAR 2 cells were transiently transfected with control vector or Myc-HRS, preincubated with cycloheximide, stimulated with PAR 2 -AP (0 -3 h), and levels of PAR 2 and ␤-actin were determined by Western blotting. Upper panels are representative blots and lower panels are densitometric analyses of PAR 2 :␤-actin. In cells expressing control vector, PAR 2 was degraded, whereas overexpression of HRS prevented degradation. *, p Ͻ 0.05. PAR 2 cells transiently transfected with control vector or HRS. In cells expressing control vector, PAR 2 -AP (120 min) caused PAR 2 trafficking to vesicles that colocalized with LAMP1 but not EEA1 (Fig. 3A). Thus, antibody-tagged PAR 2 traffics to lysosomes similarly to untagged PAR 2 (4). In contrast, HRS overexpression prevented the trafficking of PAR 2 to lysosomes and caused accumulation of PAR 2 in EEA1-containing vesicles (Fig. 3B). Quantitative analysis of these data revealed that the co-localization of PAR 2 and EEA1 was significantly increased in cells transfected with HRS compared with control vector (Fig.  3C). Conversely, the colocalization of PAR 2 and LAMP1 was significantly diminished in cells overexpressing HRS compared with control vector (Fig. 3C).
To assess the effect of HRS overexpression on the degradation of PAR 2 , we transfected HEK-PAR 2 cells with control vector or HRS. Cycloheximide-treated cells were stimulated with PAR 2 -AP (0 -3 h) and levels of PAR 2 determined by Western blotting. We found that in cells expressing control vector, PAR 2 was quickly degraded (64 Ϯ 2% of control (100%), 3 h) (Fig. 3D). In contrast, in cells overexpressing HRS, degradation of PAR 2 was prevented (96 Ϯ 7% of control, 3 h). Thus, HRS overexpression prevents the lysosomal targeting and degradation of PAR 2 and causes its retention in EEA1-containing vesicles.
Interaction of HRS and PAR 2 does not require the HRS ubiquitin-interacting motif or PAR 2 ubiquitination. HRS possesses several protein-protein interacting motifs, including an UIM (21). PAR 2 is ubiquitinated as a prerequisite to degradation (4).
To determine whether the UIM of HRS is required for interaction with PAR 2 , we expressed a mutant of HRS that lacks the UIM (HRS⌬UIM) in HEK-PAR 2 cells. In unstimulated cells, PAR 2 was at the plasma membrane and in enlarged HRS⌬UIMcontaining vesicles (Fig. 4A). PAR 2 -AP (30 -120 min) induced PAR 2 depletion from the plasma membrane and accumulation in enlarged vesicles containing HRS⌬UIM (Fig. 4A). Quantitative analysis of these data indicated an identical degree of colocalization of PAR 2 with HRS or HRS⌬UIM at 120 min after stimulation with PAR 2 -AP (Fig. 4B). To determine whether ubiquitination of PAR 2 was necessary for this interaction, we studied an ubiquitin-defective mutant of PAR 2 (PAR 2 ⌬14K/R) (4). We observed that in unstimulated cells, PAR 2 ⌬14K/R was at the cell surface and in enlarged HRS-containing vesicles (Fig.  4C). PAR 2 -AP (30 min) induced depletion of PAR 2 ⌬14K/R from the plasma membrane and accumulation in enlarged HRS-containing vesicles (Fig. 4C). Thus, the colocalization of PAR 2 and HRS is not dependent on the ubiquitin-interacting motif of HRS or on the ubiquitination of PAR 2 .
HRS overexpression prevents the ubiquitin-defective PAR 2 from recycling. The ubiquitin-defective mutant PAR 2 ⌬14K/R escapes from the normal degradative pathway and recycles to the plasma membrane (4). To examine the effect of HRS overexpression on PAR 2 ⌬14K/R recycling, we transfected HEK-PAR 2 ⌬14K/R cells with control vector or HRS, and examined endocytosis and recycling of PAR 2 ⌬14K/R labeled at the cell surface with an antibody to the extracellular FLAG epitope. In unstimulated cells expressing control vector, PAR 2 ⌬14K/R was at the plasma membrane (Fig. 5A). PAR 2 -AP (30 min) induced PAR 2 ⌬14K/R depletion from the plasma membrane and trafficking to HRS-containing endosomes. After agonist washout and recovery for 4 h, PAR 2 ⌬14K/R recycled to the plasma membrane (Fig. 5A). In unstimulated HRS-overexpressing cells, PAR 2 ⌬14K/R was at the plasma membrane and in enlarged HRS-containing endosomes (Fig. 5B), suggesting constitutive endocytosis. PAR 2 -AP (30 min) induced PAR 2 ⌬14K/R depletion from the plasma membrane and accumulation in HRS-containing endosomes. At 30 min after stimulation with PAR 2 -AP, PAR 2 ⌬14K/R and wild-type PAR 2 colocalized with HRS to the same extent (Fig. 5C). After agonist removal, PAR 2 ⌬14K/R did not recycle and was instead retained in vesicles containing EEA1 and HRS (Fig. 5B). Thus, HRS is required for post-endocytic recycling of an ubiquitin-defective PAR 2 mutant.
HRS Overexpression Prevents Resensitization of PAR 2 ⌬14K/R Ca 2ϩ Signaling-Recycling mediates resensitization of PAR 2 ⌬14K/R (4). To examine the effect of HRS overexpression on the resensitization of PAR 2 ⌬14K/R Ca 2ϩ signaling, HEK- were transiently transfected with HRS⌬UIM, stimulated with PAR 2 -AP (0, 30 or 120 min), and proteins were localized by indirect immunofluorescence using HA11 (PAR 2 ) and HRS antibodies. In unstimulated cells, PAR 2 was present at the plasma membrane (arrowheads) and colocalized with HRS⌬UIM in enlarged vesicles (arrows). In stimulated cells, PAR 2 was removed from the plasma membrane and detected in HRS⌬IUM-containing vesicles (arrows). B, quantification of the effect of deleting the UIM of HRS on colocalization of PAR 2 and HRS. The overlap coefficient between PAR 2 and HRS was measured in HEK-PAR 2 cells transiently transfected with Myc-HRS or Myc-HRS⌬UIM. n, 20 and 24 cells for Myc-HRS and Myc-HRS⌬UIM, respectively. C, HEK-PAR 2 ⌬14K/R cells were transiently transfected with Myc-HRS, and cell-surface PAR 2 labeled with an antibody to the extracellular epitope (Flag). Cells were stimulated with PAR 2 -AP (0 or 30 min) and proteins were localized by indirect immunofluorescence using fluorescent secondary antibody (PAR 2 ) and HRS antibody. In unstimulated cells, PAR 2 was present at the plasma membrane (arrowheads) and colocalized with HRS⌬UIM in enlarged vesicles (arrows). In stimulated cells, PAR 2 was removed from the plasma membrane and detected in HRS⌬UIM-containing vesicles (arrows). Scale bars, 10 m. PAR 2 ⌬14K/R cells were transfected with control vector or HRS. To prevent synthesis of new receptors or mobilization of receptors from the Golgi apparatus, which also contribute to PAR 2 resensitization (30), cells were treated with cycloheximide and brefeldin A 1 . Cells were treated with PAR 2 -AP (1 h) or vehicle (control), washed and recovered in PAR 2 -AP-free medium for 0 -4 h, and then challenged with PAR 2 -AP (100 M). When challenged without recovery, PAR 2 -AP-induced Ca 2ϩ signaling was minimal (vector control, 23 Ϯ 1% of vehicle; HRS, 22 Ϯ 2% of vehicle), indicating desensitization (Fig. 6, A  and B). However, after 4 h recovery, cells expressing control vector had resensitized (72 Ϯ 6% of vehicle). In contrast, cells overexpressing HRS showed diminished resensitization (48 Ϯ 5% of vehicle). Thus, HRS is required for the resensitization of a recycling PAR 2 ⌬14K/R mutant.
HRS Regulates Recycling and Lysosomal Trafficking of CLR, a Non-ubiquitinated GPCR-Our results with PAR 2 indicate that HRS participates in lysosomal trafficking and degradation of the wild-type receptor, and in recycling and resensitization of a non-ubiquitinated mutant, PAR 2 ⌬14K/R. We have recently shown that CLR and RAMP1 recycle after transient activation with CGRP, but traffic to lysosomes for degradation after sustained activation, and that these processes do not require ubiquitination of CLR or RAMP1 (8). We therefore examined the role of HRS in this divergent trafficking of CLR and RAMP1.
To assess whether HRS plays a role in CLR and RAMP1 trafficking, we localized endogenous HRS and CLR in CLR-RAMP1-HEK cells. In unstimulated cells, CLR was detected mainly at the plasma membrane and was not associated with HRS (Fig. 7A). CGRP (100 nM, 30 min) induced endocytosis of CLR, which co-localized with endogenous HRS (Fig. 7A). Thus, CLR internalizes and colocalizes with HRS in early endosomes, where HRS could regulate its post-endosomal trafficking.
To evaluate whether HRS mediates post-endocytic trafficking of CLR to recycling pathways, we examined the effects of overexpressing HRS on CLR recycling after transient stimulation with CGRP. Cell-surface CLR was labeled by incubating cells with an antibody to the extracellular HA11 epitope of CLR, which we have shown does not affect CLR trafficking (8). We found that in cells not overexpressing HRS, or in cells expressing the control vector, CLR was initially at the cell surface, and CGRP (30 min) induced internalization of antibody-labeled receptor, which recycled at 4 h after washing and incubating cells in CGRP-free medium (Fig. 7, B and C). In cells overexpressing HRS, CLR was initially detected at the cell surface and in some HRS-containing endosomes, suggesting constitutive endocytosis (Fig. 7C). CGRP induced depletion of CLR from the plasma FIGURE 5. Effect of HRS overexpression on recycling of PAR 2 ⌬14K/R. HEK-PAR2⌬14K/R cells were transiently transfected with control vector (pcDNA5, A) or Myc-HRS (B), and cell-surface PAR 2 labeled with an antibody to the extracellular FLAG epitope. Cells were stimulated with PAR 2 -AP (0 or 30 min), washed, and placed in PAR 2 -AP-free medium for 4 h. Proteins were localized by indirect immunofluorescence using fluorescent secondary antibody (PAR 2 ) and HRS and EEA1 antibodies. A, in unstimulated cells expressing control vector, PAR 2 ⌬14K/R was present at the cell surface and HRS was present in intracellular vesicles. In stimulated cells, PAR 2 ⌬14K/R was detected in HRS-containing vesicles (arrows). In recovered cells, PAR 2 ⌬14K/R was recycled to the cell surface (arrowheads), and a minor proportion was present in HRS-and EEA1-containing endosomes (arrows). B, in unstimulated cells expressing Myc-HRS, PAR 2 ⌬14K/R was at the cell surface (arrowheads) and in enlarged HRS-containing intracellular vesicles (arrows). In stimulated cells, PAR 2 ⌬14K/R was detected in HRS-containing vesicles (arrows). In recovered cells, PAR 2 ⌬14K/R did not return to the cell surface and was retained in HRS and EEA1-containing endosomes (arrows). Scale bars, 10 m. C, quantification of the effect of deleting all intracellular lysine residues of PAR 2 on colocalization of PAR 2 and HRS. The overlap coefficient between PAR 2 and HRS was determined in HEK-PAR 2 and HEK-PAR 2 ⌬14K/R cells both transiently transfected with Myc-HRS. n, 22 and 23 cells for PAR 2 and PAR 2 ⌬14K/R, respectively. membrane; CLR was then retained in HRS-containing, EEA1positive endosomes and did not recycle after 4 h (Fig. 7C). Thus, disrupting the function of HRS by overexpression of HRS blocks CLR recycling.
To evaluate whether HRS mediates post-endocytic trafficking of CLR to degradative pathways, we examined the effects of overexpressing HRS on CLR trafficking to lysosomes after sustained stimulation with CGRP. In cells not overexpressing HRS, or in cells expressing the control vector, CGRP (4 h) induced trafficking of antibody-labeled CLR to vesicles containing LAMP1 but not EEA1 (Fig. 7D). In cells overexpressing HRS, CLR was retained in HRS-containing endosomes, and was not detected in LAMP1-containing lysosomes after 4 h (Fig. 7D). Thus, disrupting HRS blocks lysosomal trafficking of CLR.
The VHS Domain of HRS Is Required for HRS-dependent Recycling of CLR but Not PAR 2 ⌬14K/R-When overexpressed, a truncation mutant of HRS lacking the N-terminal VHS domain is defective in its ability to inhibit recycling of the ␤ 2 -adrenergic receptor, suggesting that the VHS domain is required for controlling recycling of this receptor (17). To examine whether the VHS domain of HRS is also required for HRS-dependent recycling of CLR and PAR 2 ⌬14K/R, we expressed a mutant of HRS lacking the VHS domain (HRS⌬VHS) in HEK-CLR and HEK-PAR 2 ⌬14K/R cells. In unstimulated cells, PAR 2 ⌬14K/R and CLR were at the plasma membrane and in some HRS⌬VHS-containing vesicles (Fig. 8,   A and B). After stimulation with agonist for 30 min, PAR 2 ⌬14K/R and CLR were depleted from the plasma membrane and accumulated in enlarged vesicles containing HRS⌬VHS (Fig. 8, A and B). After agonist removal and recovery for 4 h, CLR recycled to the plasma membrane, with a portion still retained in HRS⌬VHS-containing vesicles (Fig. 8B). In contrast, PAR 2 ⌬14K/R did not recycle and was completely retained in HRS⌬VHS-containing vesicles (Fig. 8A). Quantification of the plasma membrane targeting after 4 h of recovery revealed that overexpressed HRS⌬VHS was as potent as wild-type HRS in preventing recycling of PAR 2 ⌬14K/R, whereas HRS⌬VHS was significantly less effective than wild-type HRS in preventing recycling of CLR (Fig. 8C). Thus, the VHS domain of HRS is required for recycling of CLR but not PAR 2 ⌬14K/R.
Knockdown of HRS Inhibits Recycling and Lysosomal Trafficking of PAR 2 and CLR-Knock-down of endogenous HRS by transfection of siRNA disrupts function similar to HRS overexpression (14,17). To determine whether our findings were an indirect consequence of HRS overexpression, we examined the effects of depleting cellular HRS on post-endocytic trafficking of PAR 2 and CLR. We chose HeLa cells for these experiments because a substantial reduction in cellular HRS can be achieved in this cell line by transfection with HRS-specific siRNA but not control (non-silencing) siRNA (14,34,35). When HeLa cells were transfected with siRNA to two distinct regions of HRS (HRS-siRNA1 or HRS-siRNA2), there was a Ͼ70% reduction in HRS levels detected in Western blots of whole cell lysates when compared with untransfected cells or cells transfected with control siRNA (Fig. 9A). The effects of HRS-siRNA1 and HRS-siRNA2 on receptor trafficking were identical. HRS-siRNA strongly inhibited recycling of transiently activated PAR 2 ⌬14K/R and CLR. Thus, after stimulation with agonist for 30 min and 4 h recovery, PAR 2 ⌬14K/R and CLR were recycled to the plasma membrane in cells expression control siRNA, but were retained in enlarged EEA1-positive endosomes in cells expressing HRS-siRNA (Fig. 9, B and C). Quantification of plasma membrane targeting after 4 h of recovery revealed that HRS-siRNA inhibited recycling of both PAR 2 ⌬14K/R and CLR (Fig. 9D).
To evaluate whether HRS knockdown affects trafficking of PAR 2 and CLR to degradative pathways, we examined the effects of HRS-siRNA on PAR 2 and CLR trafficking to lysosomes after sustained stimulation with agonist. After continuous stimulation with agonist for 4 h, PAR 2 and CLR were detected in LAMP1-containing lysosomes in cells expressing control siRNA, but were retained in enlarged EEA1-containing endosomes in cells expressing HRS-siRNA (Fig. 9, E and  F). Thus, consistent with or findings of HRS overexpression, HRS knockdown results in retention of internalized PAR 2 , PAR 2 ⌬14K/R, and CLR in enlarged EEA1-containing endosomes, thereby preventing the post-endocytic sorting of these receptors to recycling or lysosomal pathways. Moreover, HRS has these divergent functions in different cell lines (HEK and HeLa).

DISCUSSION
We have identified two distinct functions of HRS in the postendocytic sorting of a single receptor, PAR 2 . First, HRS medi- ates agonist-induced trafficking of ubiquitinated PAR 2 to lysosomes, where degradation occurs. Second, HRS mediates recycling and resensitization of non-ubiquitinated PAR 2 . We have also shown that HRS has a similar role in trafficking CLR, which is not ubiquitinated after activation. HRS mediates lysosomal trafficking of CLR after sustained activation, and recycling after transient activation. The role of HRS in the sorting of a single receptor into recycling and degradative pathways has not been demonstrated before. This study on PAR 2 and CLR trafficking therefore provides a unique opportunity to dissect the role of HRS in these divergent sorting pathways and evaluates the role of receptor ubiquitination in this process.
HRS-dependent Trafficking of Ubiquitinated PAR 2 and Its Ubiquitin Ligase c-Cbl-Although c-Cblmediated ubiquitination of PAR 2 is required for sorting of the receptor from early endosomes to lysosomes, the mechanisms of this ubiquitindependent sorting are unknown. We observed that disruption of HRS function by overexpression prevented agonist-induced lysosomal trafficking and degradation of PAR 2 . This inhibition was due to the accumulation of PAR 2 in enlarged HRS-containing endosomes. Thus, we have established a critical role for HRS in the lysosomal trafficking and subsequent degradation of PAR 2 . These findings are consistent with studies demonstrating that HRS mediates lysosomal targeting of other ubiquitinated GPCRs, such as CXCR4 (16). HRS recognizes ubiquitinated proteins through its UIM, and thereby functions as an adaptor between ubiquitinated cargo and the endosomal sorting machinery that is responsible for multivesicular body formation and receptor sorting to lysosomes (21,39). The overexpression of HRS disrupts inward vesiculation during multivesicular body formation and receptor sorting to lysosomes, indicating a role for HRS in both of these processes (40). Ubiquitinated receptors, such as epidermal growth factor receptor, interact with HRS in the multivesicular body and are targeted to lysosomes, whereas non-ubiquitinated receptors, such as TfR, may diffuse into this region but are not retained within it and recycle to the plasma membrane (20).
The E3 ligase c-Cbl is responsible for the ubiquitination of PAR 2 and translocates to the plasma membrane after activation of the receptor and traffics with the receptor to early endosomes before returning to the cytosol (4). We observed that after disrupting HRS function and activating PAR 2 , c-Cbl traf- , and proteins were localized by immunofluorescence using HA11 (CLR) and HRS antibodies. In unstimulated cells, CLR was present at the cell surface (arrowheads) and HRS was present in intracellular vesicles (arrows). In stimulated cells, CLR was depleted from the plasma membrane, and was detected in HRS-containing endosomes (arrows). B and C, HEK-CLR-RAMP1 cells were transiently transfected with control vector (pcDNA5) or Myc-HRS, and cell-surface CLR was labeled with an antibody to the extracellular epitope (HA11). Cells were stimulated with CGRP (0 or 30 min), washed, and placed in CGRP-free medium for 0 -4 h. Proteins were localized using fluorescent secondary antibody (CLR) and HRS and EEA1 antibodies. B, in unstimulated cells expressing control vector, CLR was present at the cell surface (arrowheads). In stimulated cells, CLR was depleted from the plasma membrane and detected in intracellular vesicles (arrows). In recovered cells, CLR was recycled to the cell surface (arrowheads). C, in unstimulated cells expressing Myc-HRS, CLR was present at the cell surface (arrowheads) and in HRS-containing vesicles (arrows). In stimulated cells expressing Myc-HRS, CLR was depleted from the plasma membrane and detected in HRS-containing vesicles (arrows). In recovered cells expressing Myc-HRS, CLR was still retained in HRS-containing vesicles (arrows). In contrast, in cells expressing control vector, CLR trafficked back to the cell surface following stimulation and recovery (arrowheads). In cells expressing Myc-HRS, CLR was retained in EEA1-containing endosomes (arrows) after recovery. D, HEK-CLR-RAMP1 cells were transiently transfected with control vector (pcDNA5) or Myc-HRS, and cell-surface CLR was labeled with an antibody to the extracellular epitope (HA11). Cells were stimulated with CGRP (4 h), and proteins were localized using fluorescent secondary antibody (CLR) and HRS and LAMP1 antibodies. In stimulated cells expressing control vector, CLR colocalized with LAMP1 in lysosomes (arrows). In cells expressing Myc-HRS, CLR did not colocalize with LAMP1 but did colocalize with HRS (arrows). Scale bars, 10 m. fics normally to the plasma membrane but accumulates with PAR 2 in HRS-containing endosomes. Similar trafficking of AIP4, the E3 ligase responsible for the ubiquitination of CXCR4, has been observed (16). Our observation raises the possibility that, similarly to AIP4, c-Cbl may mediate ubiquitination of HRS on endosomes, which could influence the function of HRS in the sorting of PAR 2 . Indeed, c-Cbl, which is also responsible for the ubiquitination of the epidermal growth factor receptor, enhances HRS ubiquitination, thereby altering the composition of the endosomal sorting machinery and its ability to target this receptor for lysosomal degradation (18).
Ubiquitin-independent Interaction between HRS and PAR 2 -The ubiquitin-defective PAR 2 mutant PAR 2 ⌬14K/R is expressed at the plasma membrane, signals normally, yet upon activation, it escapes normal lysosomal targeting and recycles (4). To investigate if ubiquitination of the receptor or the UIM of HRS were necessary for the interaction between PAR 2 and HRS, we studied the trafficking of PAR 2 and PAR 2 ⌬14K/R in the presence of overexpressed HRS⌬UIM and HRS, respectively. In both cases, the receptor accumulated in enlarged HRS-containing endosomes. Thus, HRS-mediated post-endocytic trafficking of PAR 2 does not require receptor ubiquitination or the UIM of HRS. These findings are consistent with the report that HRS-mediated lysosomal trafficking of ␦-opioid receptor is independent of receptor ubiquitination (14).
New receptor synthesis and transport of Golgi stores to the plasma membrane mediate resensitization of wild-type PAR 2 (30). PAR 2 ⌬14K/R can resensitize by recycling (4), which provided an opportunity to determine the role of HRS in this process. We observed that overexpression of HRS prevented recycling and resensitization of activated PAR 2 ⌬14K/R. In support of our results, HRS also mediates recycling of both the ␤ 2 -adrenergic receptor and the -opioid receptor, which is critical for resensitization of signaling (17).
Role of HRS in Post-endocytic Sorting of CLR, a Naturally Nonubiquitinated GPCR-By examining the post-endocytic sorting of CLR, we were able to define the contribution of HRS to recycling and lysosomal trafficking of a non-ubiquitinated GPCR (8). We observed that activated CLR traffics to endosomes containing endogenous HRS, and then either recycles (after transient stimulation with CGRP) or traffics to lysosomes (after sustained stimulation). Overexpression of HRS blocked both recycling and lysosomal trafficking of CLR, and presumably RAMP1, by trapping the receptor in HRS-positive enlarged endosomes. Because CLR and RAMP1 are not ubiquitinated after activation, these results support the role of HRS in post-endocytic sorting of non-ubiquitinated GPCRs to recycling and lysosomal pathways (14,17).
The observation that PAR 2 , PAR 2 ⌬14K/R, and CLR all accumulate in HRS-containing endosomes after activation could be due to a nonspecific effect of HRS overexpression on endosomal function. To investigate this possibility, we examined the effects of HRS disruption on trafficking of TfR, which constitutively endocytoses and recycles (21). Notably, overexpression of HRS did not affect TfR trafficking, which is in support of previous reports of HRS-independent endocytosis and recycling of FIGURE 8. Effects of HRS⌬VHS overexpression on recycling of PAR 2 ⌬14K/R and CLR. A, HEK-PAR 2 ⌬14K/R cells were transiently transfected with Myc-HRS⌬VHS and cell-surface PAR 2 ⌬14K/R labeled with an antibody to the extracellular FLAG epitope. Cells were stimulated with PAR 2 -AP (0 or 30 min), washed, and placed in PAR 2 -AP-free medium for 4 h. Proteins were localized by indirect immunofluorescence using fluorescent secondary antibody (PAR 2 ) and HRS antibody. In unstimulated cells, PAR 2 ⌬14K/R was at the cell surface (arrowheads) and HRS⌬VHS was in intracellular vesicles. In stimulated cells, PAR 2 ⌬14K/R was detected in HRS⌬VHScontaining vesicles (arrows). In recovered cells, PAR 2 ⌬14K/R remained in HRS⌬VHS-containing vesicles (arrows) and did not recycle to the cell surface. B, HEK-CLR-RAMP1 cells were transiently transfected with Myc-HRS⌬VHS and cell-surface CLR labeled with an antibody to the extracellular HA11 epitope. Cells were stimulated with CGRP (0 or 30 min), washed, and placed in CGRP-free medium for 4 h. Proteins were localized by indirect immunofluorescence using fluorescent secondary antibody (CLR) and HRS antibody. In unstimulated cells, CLR was at the cell surface (arrowheads) and HRS⌬VHS was in intracellular vesicles. In stimulated cells, CLR was detected in HRS⌬VHS-containing vesicles (arrows). In recovered cells, CLR was recycled to the cell surface (arrowheads), and a proportion was present in HRS⌬VHS-containing vesicles (arrows). Scale bars, 10 m. C, quantification of the effects of Myc-HRS and Myc-HRS⌬VHS on the recycling of PAR 2 ⌬14K/R and CLR. The percentage of total fluorescence at the plasma membrane after 4 h of recovery was determined. n, 20 to 30 cells each. *, p Ͻ 0.05.
for the function of HRS in PAR 2 ⌬14K/R recycling. The VHS domain of HRS may thus have an important role in linking some but not all GPCRs to HRS, which requires further investigation.
HRS Knockdown Has the Same Inhibitory Effects on Postendocytic Trafficking of PAR 2 and CLR as HRS Overexpression-Depletion of endogenous HRS using siRNA produces an enlargement of EEA1-positive endosomes identical to overexpression (14,16,17,24,41). We noted that the enlargement of EEA1-positive endosomes was more pronounced in HEK cells compared with HeLa cells, which is consistent with previous studies (16,17). The effects of HRS knockdown and HRS overexpression on the trafficking of PAR 2 and CLR in either HEK or HeLa cells were identical. Disrupting HRS function by either approach prevented recycling of PAR 2 ⌬14K/R and CLR after agonist washout and recovery, and inhibited lysosomal trafficking of PAR 2 and CLR during prolonged stimulation. HRS knockdown and overexpression have been previously reported to have inhibitory effects on trafficking of other receptors, such as the ␤ 2 -adrenergic receptor, -opioid receptor, and epidermal growth factor receptor (14,17,34,44).
In summary, the present results demonstrate for the first time the critically important role of HRS in the divergent postendocytic sorting of individual GPCRs to both degradative and recycling pathways, which have opposite functional effects on cell signaling. HRS is required for lysosomal trafficking and degradation of PAR 2 and CLR, which would prevent sustained signaling of these receptors that could otherwise result in uncontrolled pain and inflammation. HRS also mediates recycling and resensitization of PAR 2 ⌬14K/R and transiently activated CLR. These sorting roles of HRS are independent of its ubiquitin-interacting motif, and HRS can regulate trafficking of ubiquitinated PAR 2 and non-ubiquitinated PAR 2 and CLR. Thus, the ubiquitin-dependent sorting decision that leads to degradative or recycling pathways appears to be downstream from HRS and requires further investigation.